Correlation Analysis between Dietary Nutrients and Fecal Bacterial Community

Distance-based redundancy analysis (db-RDA) (Legendre and Anderson, 1999) was performed to correlate macronutrient intake with fecal bacterial composition. The individual datasets (n = 24 for Baybay; n = 16 for Ormoc) of macronutrient energy contributions and family level bacterial composition were subjected to the ‘capscale’ function in the R vegan package⁶. The Bray–Curtis distance was used to coordinate bacterial community variation. The covariable effect between dietary nutrients and city of residence was estimated by the conditioning function in the ‘capscale’ function. Values of R² and p were estimated by the ‘RsquareAdj’ and ‘Anova’ functions in the R vegan package, respectively.

We also analyzed the correlation between each dietary nutrient and fecal bacterial community using the ‘envfit’ function in the R vegan package (Oksanen, 2015). The correlation was calculated using individual macro- and micronutrient intakes and family level PCA data. The correlation between fat intake and the relative abundance of each bacterial taxonomic group was calculated in Stata SE12.0 by Spearman’s rank correlation test, using data on the fat energy ratio and relative abundance of each taxonomic group.

To create an association network map among bacteria species and three macronutrients, we calculated pairwise Spearman’s correlation coefficients between species and between species and each macronutrient, according to the relative abundance of species and energy ratio of three macronutrient of Leyte children. The Spearman’s correlation coefficients (rho) were obtained with p-value by using the out.association program in Mothur software ver. 1.35.1⁷ (Schloss et al., 2009). Correlation coefficients higher than 0,4 or lower than -0.4 with p < 0.05 were extracted and visualized as edges connecting two species nodes in the Prefuse Force Directed Layout using Cytoscape 3.3.0⁸ (Lopes et al., 2010).